Generative AI chatbots are supplying inaccurate information about a majority of UK retailers, creating a new source of customer confusion over locations, opening details, products, and brand identity.
Research by AI visibility platform Searchable found that 64% of the retailers tested were affected by at least one false statement across ChatGPT, Gemini, and Perplexity.
More than 72,000 questions were tested against verified company information. The systems attempted to answer 98% of queries, but one in 16 responses contained an error.
Inaccuracies included practical details such as addresses, contact information, and store locations. Some incorrect postcodes could have directed a customer more than 20 miles from the intended destination.
Smaller retailers were more exposed than companies with extensive digital footprints. Businesses with fewer authoritative webpages, inconsistent directory entries, or limited third party coverage provide AI systems with less reliable material from which to construct an answer.
The findings identify a customer experience risk outside many retailers’ conventional digital controls. A company can correct its website and official listings, but it cannot directly determine the wording produced by an independent chatbot.
Consumers are increasingly asking conversational systems for product recommendations, opening hours, nearby stores, returns information, and comparisons. Because the response is often presented as a direct answer rather than a list of links, users may not visit the retailer’s website to check whether the information is accurate.
An incorrect search listing is usually traceable to a particular profile or webpage. Generative answers draw from several sources and can vary when the same question is repeated, making errors harder to reproduce, diagnose, and correct.
The problem extends well beyond retail. Hotels, restaurants, professional services companies, healthcare providers, venues, and local trades all depend on accurate information about location, availability, and services. Organisations with several branches or frequently changing offers face an especially demanding maintenance task.
Information consistency will need to form part of customer experience management rather than remaining a narrow search marketing responsibility. Official websites, store finders, product feeds, business profiles, structured data, and major directories should all contain matching details.
Fragmented information creates similar problems inside organisations. An earlier examination of the data weaknesses affecting AI enabled loyalty programmes showed how conflicting customer records can limit personalisation. Public inconsistencies now risk weakening the answers customers receive before entering any company controlled channel.
Ownership of the underlying information is often divided. Store operations may control opening hours, property teams may manage addresses, marketing departments may oversee search profiles, and technology teams may maintain structured website data. Corrections made in one system can leave outdated details elsewhere.
Monitoring practices will consequently need to expand. Traditional brand tracking concentrates on media coverage, search rankings, reviews, and social discussion, whereas conversational search requires companies to test how leading systems answer common customer questions.
Such testing must account for the probabilistic nature of generative AI. A single accurate response does not demonstrate that every user will receive the same answer, so reviews should cover several platforms, question formats, and locations.
AI providers also face questions about responsibility. A confidently stated but incorrect address or service description can cause inconvenience and lost revenue. Clearer attribution, visible sources, and effective correction mechanisms would allow users and companies to challenge mistakes more easily.
Consumer protection concerns become more serious where inaccurate answers affect prices, financial decisions, availability, or access to essential services. Regulators may eventually consider whether certain categories of commercial information require stronger safeguards.
Improving public data is the most immediate response available to retailers. Accurate structured information, consistent naming, current branch details, and authoritative location pages can reduce ambiguity, although they cannot guarantee correct chatbot output.
Smaller companies face a resource disadvantage because they are less likely to employ dedicated search, data, or digital governance specialists. Trade bodies, technology suppliers, and local business networks may therefore need to provide practical guidance as conversational search becomes more influential.
The commercial effect of each error will vary. An incorrect opening time may produce a wasted journey, while a false statement about returns or product availability could lead a customer to choose a competitor without contacting the retailer.
Repeated inaccuracies can also damage trust even where the retailer did not produce them. Consumers may not distinguish clearly between information published by a brand and an answer generated by an external system.
Retailers have spent years reducing friction across websites, applications, stores, and contact centres. Conversational search has introduced another customer interface, operated by technology platforms but influenced by the quality and consistency of the information available to them.
The research does not establish how frequently incorrect answers become lost transactions, and model behaviour will continue to change. It does show that confident responses cannot be treated as reliable simply because they are fluent, detailed, or presented without visible uncertainty.
You must be logged in to post a comment.